Prospect Theory and Why People Resist New Systems Even When They Are Better

There is a pattern I keep seeing in IS implementation projects. A new system arrives. Objectively, it is faster, more capable, and better integrated than the old one. The vendor has benchmarks. The IT team has a business case. And still, a large portion of users resist it, complain about it, or find creative ways to avoid using it the way it was designed. The standard explanation is change management. People just do not like change.

But that explanation is too thin. It does not tell you why the resistance exists at a psychological level, or why it is often stronger than the objective improvement warrants. Kahneman and Tversky's prospect theory, developed in 1979, gives a sharper account.

The core of the theory is the value function: losses hurt more than equivalent gains feel good. The original Kahneman and Tversky (1979) paper is cited in the study-hub materials I work from, specifically in papers studying shareholder reactions to negative events, where the loss aversion explanation predicts asymmetric responses to gains and losses. The mechanism is the same regardless of domain. People are not symmetric in how they evaluate outcomes. Losing $100 feels worse than gaining $100 feels good, by roughly a factor of two in the classic formulation. This asymmetry has been shown across a wide range of contexts.

A second, equally important part of the theory is reference dependence. People do not evaluate outcomes in absolute terms. They evaluate them relative to a reference point, which is usually the current state. The old system is the reference point. The new system is not evaluated on its own merits. It is evaluated as a bundle of changes from the reference point, some of which are gains and some of which are losses, and the losses are weighted more heavily.

When I think about system migration through this lens, the math changes. What feels like a straightforward improvement to the people who designed it looks very different to the people who have to live through the transition. Users with the old system face specific, concrete, certain losses, re-learning keyboard shortcuts and workflows they built over years, disruption to their daily patterns during the transition period, the loss of small contextual features and workarounds that made their particular job easier, and the social loss of no longer being the expert who knows where everything is. These losses are immediate and guaranteed.

The gains, on the other hand, are uncertain and delayed. The new system promises better reporting, faster processing, and improved integration. But those gains are not felt on day one. They require learning, which takes time and produces errors during the learning period. A user who is slower and making more mistakes during the first month of a new system is experiencing that as real loss, even if by month six the performance gains materialize. Prospect theory predicts that the asymmetric weighting of immediate certain losses against delayed uncertain gains will produce resistance even when the gains are genuine.

This is different from the explanation that users are irrational or poorly trained. The resistance is a predictable response to an asymmetric cost-benefit structure. The question is not how to make users more rational, but how to restructure the adoption experience so the cost-benefit profile looks different.

I think there is a practical implication buried here that most implementations miss. The standard response to resistance is to do more communication, more training, and more executive sponsorship messages. These are aimed at the perceived usefulness and the information gap. They treat resistance as a knowledge problem or an attitude problem. But if the real driver is loss aversion, then the interventions need to target the reference point and the loss structure, not just the perception of the new system's features.

One way to do this is to make the gains more immediate and less uncertain, and to make the losses feel smaller by not eliminating familiar elements all at once. Gradual migration, preserving familiar workflows in a parallel mode during transition, giving users early wins that they can feel rather than only read about in a vendor benchmark, these are loss aversion interventions even if they are not called that. The goal is to change the reference point structure so users are not facing a big bundle of guaranteed losses upfront.

Another implication is about who bears the losses in a migration. In most enterprise implementations, the people who bear the most disruption during transition are not the same people who authorized the investment or will see the cost savings on a finance report. The users doing data entry or customer service calls are losing productivity and learning time. The executives see the aggregate efficiency number. Prospect theory would predict that the people carrying the losses will weight them heavily, while the people who will eventually see the aggregate gain on a spreadsheet will not adequately account for the loss experience.

Gartner has consistently noted in its research on digital initiatives that technology selection is rarely the primary failure point, while user adoption and change management are where projects most commonly fall short (see Gartner newsroom for research on digital initiative success factors). My read of this pattern, filtered through prospect theory, is that adoption problems are not mainly about awareness or attitude. They are about the loss structure of the transition, and the fact that implementation designs consistently front-load the costs and back-load the benefits.

The reference dependence point also matters for how you sequence migrations. If users go through a big painful transition and then the next system replacement arrives before they have fully recovered their prior productivity level, their new reference point is still negative from the first transition. The second wave of losses compounds on an already-below-baseline starting point. Organizations that do serial migrations without enough recovery time between them are essentially running a prospect-theory nightmare.

Kahneman's broader work, including the book Thinking, Fast and Slow (2011), develops these ideas across many domains. The core insight, that losses loom larger than gains, applies wherever people evaluate changes from a status quo. For IS researchers, the implication is that models like TAM, which focus on perceived usefulness and ease of use, may be missing the loss-weighting dynamic. A system can score well on perceived usefulness and still face strong resistance if users are evaluating it against a reference point that makes the transition look like a net loss, even when the long-run math is positive.

The users resisting your new system are probably not wrong about what they are losing. They are just weighting it the way humans weight losses, which is heavily, and the implementation design probably did not account for that.